biblio.ugent.be The UGent Institutional Repository is the electronic archiving and dissemination platform for all UGent research publications. Ghent University has implemented a mandate stipulating that all academic publications of UGent researchers should be deposited and archived in this repository. Except for items where current copyright restrictions apply, these papers are available in Open Access. This item is the archived peer-reviewed author-version of: Title: Real-time assessment of critical quality attributes of a continuous granulation process Authors: M. Fonteyne, J. Vercruysse, D Córdoba Diaz, D. Gildemyn, C. Vervaet, J.P. Remon, T. De Beer In: Pharmaceutical Development and Technology, 1-13 (2011) Optional: link to the article To refer to or to cite this work, please use the citation to the published version: Authors (year). Title. journal Volume(Issue) page-page. Doi 10.3109/10837450.2011.627869
42
Embed
Title: Real-time assessment of critical q uality attributes of a … · Lindberg et al. [11 ] used a twin screw extruder to granulate. Both powder flow rate and liquid flow rate were
This document is posted to help you gain knowledge. Please leave a comment to let me know what you think about it! Share it to your friends and learn new things together.
Transcript
biblio.ugent.be
The UGent Institutional Repository is the electronic archiving and dissemination platform for all UGent research publications. Ghent University has implemented a mandate stipulating that all academic publications of UGent researchers should be deposited and archived in this repository. Except for items where current copyright restrictions apply, these papers are available in Open Access.
This item is the archived peer-reviewed author-version of:
Title: Real-time assessment of critical quality attributes of a continuous granulation process
Authors: M. Fonteyne, J. Vercruysse, D Córdoba Diaz, D. Gildemyn, C. Vervaet, J.P. Remon, T. De Beer
In: Pharmaceutical Development and Technology, 1-13 (2011)
Optional: link to the article
To refer to or to cite this work, please use the citation to the published version:
Authors (year). Title. journal Volume(Issue) page-page. Doi 10.3109/10837450.2011.627869
Real-time assessment of critical quality attributes of a continuous
µm and the particle size distribution was calculated in real-time every five seconds. A shortfall of the
Parsum system is that once the particle size distribution ranges are chosen, they cannot be adjusted
during the process. This means that prescience on the particle size distribution of the investigated
granules is required before starting the SFV measurements. All measurement values were
continuously stored and visualized in real-time by the IPP 70 software. Data analysis was done using
the V 1.6 Macro in Excel 2007.
Design of Experiments (DOE)
A 2-level full factorial design was applied to study the influence of four process variables upon both
the solid state and particle size (D50) of the wet granules (i.e. 2^4 + 3 = 19 granulations
experiments). The screw speed was varied between 600 and 950 rpm. The temperature of the
granulator barrel was altered between 25°C and 40°C. The liquid concentration was varied between
8.38% and 9.94% (w/w). The premixed powder was fed in a range from 10 to 25 kg per hour (Table
2). Both the powder feed rate and screw speed determine the barrel filling degree. The center point
experiment of the design was repeated three times. All equipment variables (Table 1) were kept
constant. The excipients, API and granulation liquid were the same for all experiments. All
performed DOE granulation experiments are listed in Table 2.
Data-analysis
The results of the DOE granulation experiments were analyzed using Modde 9.0 (Umetrics, Umeå,
Sweden). Simca P+ 12.1 (Umetrics, Umeå, Sweden) was used for principal component analysis (PCA)
of the NIR and Raman spectra collected during all granulation experiments. The spectral data were
Standard Normal Variation (SNV) corrected for baseline-correction and for neutralizing differences in
particle size and packing of the granules before applying PCA.
Results and Discussion
One characteristic of the ConsiGma™-25 is that all continuously produced wet granules are
transferred to the fluid bed dryer. Granules which are not meeting the predetermined specifications
cannot be removed, as is possible for batch production systems. Since batches are inspected after
production and off-line tests are performed, bad batches can be rejected and will not be processed
to the next production step. Therefore knowledge and control of the critical characteristics of the
granules is important to guarantee appropriate downstream processing.
Particle size
The in-line particle size analyzer (Parsum®, Chemnitz, Germany) gives a very rapid image of the
particle size distribution of the continuously produced wet granules. However, during the DoE
experiments, sticking of the wet granules to the probe window (fouling) was observed despite the
integrated air-cleaning system. Therefore, only these distributions in which no fouling was seen were
extracted from all obtained distributions per DoE experiment (every 5 s during 3 minutes) for further
interpretation. For future studies and more detailed wet granule size distribution evaluation, we will
first need to examine whether fouling can be avoided by optimizing the probe interfacing and
cleaning system. Therefore in this study, only the D50 values from the selected distributions were
evaluated to observe some trends how the examined parameters influence the granule size. A
representative example of a selected particle size distribution obtained during DoE experiment 4 is
shown in Figure 3. The D50 (50 % of the distribution has a particle size smaller than this value)
obtained after two minutes of measurements was used as the DOE-response value. Table 2 gives an
overview of the D50’s after two minutes of production for all DOE-experiments. The influence of the
process variables upon the D50 is evaluated in the effect plot, shown in Figure 4. Both the powder
feed rate and the temperature of the barrel have a significant and positive influence on the D50.
Increasing the powder feed rate and increasing the barrel temperature result in large and oversized
granules. Also, a significant interaction between these two factors was found. A higher powder feed
rate will result in a higher compaction of the powder in the granulator barrel and thus larger
granules. A higher temperature of the granulator barrel will induce a higher solubility of lactose,
theophylline and PVP and consequently more bridge formation. Adding more granulation liquid
tends to generate larger granules, but only a small range of the amount of added liquid was
investigated in this study. Experiments 2, 9 and 19 illustrate these findings, as these three runs
produced very large granules (Table 2). Experiments 1, 4, 6, 12 and 18 produced the smallest
granules, as granulation was done using a low powder feed rate (10 kg/hour) and (except for run 18)
a cold granulator barrel.
Overall it can be stated that none of the produced wet granules are satisfactory regarding particle
size. The produced granules are too large to obtain good tablets after compression, but some
remarks should be made. First off all, this is a measurement of wet granules, the particle size of the
granules may reduce during drying. Furthermore the ConsiGma™-25 is equipped with a discharging
unit, which contains a mill to grind oversized granules. Nevertheless, the aim should be to fully
understand, monitor and control the granulation and drying process, so that granules with an ideal
particle size distribution are obtained and the mill will be redundant.
Solid State
Raman spectroscopy
Wikström et al. [27] found that Raman spectroscopy is an efficient tool to differentiate theophylline
monohydrate from anhydrous theophylline, and that the conversion can be followed in real-time.
The spectrum (Figure 5) of TA has two intense peaks at 1664 and 1706 cm-1, whereas TH has a peak
at 1686 cm-1, which is due to C=O stretch of the carbonyl-groups. Furthermore during hydrate
formation the O==C—N band shifts from 555 cm-1 to 572 cm-1 [29]. The excipients in the studied
formulation have no overlapping Raman signal in these regions.
Principal component analysis of all spectra collected during the 19 DOE experiments was performed.
The spectral regions from 520 cm-1 to 600 cm-1 and from 1636 cm-1 to 1737 cm-1 were selected and
modeled together, as the solid state information is available in these spectral regions. Three
principal components described 99.4% the spectral variation. The first, second and third PC
explained 78.62%, 14.37% and 6.41%, respectively. A PC 1 versus PC 2 scores plot was plotted, and
Gewijzigde veldcode
information regarding the solid state can be seen along the PC2-axis. Secondary observation
identifiers (i.e., the applied colors in the scores plots) were assigned to each individual spectrum
(score) in order to classify them after used screw speed, barrel temperature, liquid concentration
and powder feed rate. When classifying spectra according to the amount of added granulation liquid
(Figure 6a; the red triangles correspond to the spectra collected from the DOE experiments where
8.38% (w/v) granulation liquid was used, the green crosses to 9.16% (w/v) and the blue squares to
9.94 % (w/v)), clustering can be noticed in the scores plot. Granules, which were produced with a
low concentration of granulation liquid, are mainly situated in the positive part of the PC2, whilst
those granulated with a high amount of liquid are situated in the negative part. Classification
according to the applied barrel temperature (Figure 6b) shows, that granules produced in a barrel
heated to 40°C result in spectra mainly clustered in the positive half of the PC 2. Granules produced
using a cold barrel (25°C) can be found in the negative part of PC 2. Concerning the powder feed
rate, the positive part of the PC2 contains most of the runs, where a powder feed rate of 25 kg/hour
was used (Figure 6c). When data were classified after screw-speed, no clustering could be seen in
the scores plot.
To find an explanation for these clusterings, one should take a look at the loadings plots. No solid-
state clustering can be seen along the PC1-axis from the scores plots, but only along the PC2-axis.
This indicates that the variation in solid state is expressed by PC 2. The loadings plot of the first
principal component shows basically the joint spectra of theophylline monohydrate and anhydrous
theophylline. The variation captured by PC 1 is most probable due to altering measurement
conditions between the different analyzed samples (e.g., differences in probe to sample distances,
differences in physical properties and granule sizes of the different samples etc.). More interesting is
the loadings plot of the second principal component (Figure 7). Maxima are found at 555 cm-1,
1665.6 cm-1 and 1707.6 cm-1, whereas minima are found at 574.2 cm-1 and 1687.5 cm-1. Both minima
can be attributed to theophylline monohydrate, while the three maxima can be assigned to
anhydrous theophylline. Hence, experiments where the spectra are clustered in the positive part of
the PC2-axis contain a quantity of remaining anhydrous theophylline. These experiments are listed in
Table 3.
It is clear that water, and a certain water activity is needed for the conversion of anhydrous
theophylline to theophylline monohydrate. Hence, the presence of remaining anhydrous
theophylline can be explained for the runs with lower liquid concentration. Adding less water, which
results in a lower water activity, leads to a lower conversion rate of anhydrous theophylline to
theophylline monohydrate [30, 31].
Ticehurst et al. [30] stated that at higher temperature an increased water activity is required to
transform anhydrous theophylline to theophylline monohydrate. This can explain why runs with a
higher barrel temperature result in these small remaining amounts of anhydrous theophylline in the
granules. Furthermore hydration is an exothermic process, which means adding energy by means of
heat will shift the reaction towards the anhydrate form [32].
With a high barrel filling the granulation liquid will have more difficulties to spread well through the
powder bed, this can be an explanation of remaining anhydrate when granulating at 25 kg/h.
Furthermore a high barrel filling will result in larger granules (as stated earlier), which makes it
harder for the granulation liquid to penetrate in the granules.
Raman spectroscopy together with Design of Experiments showed to be valuable tools to investigate
and understand solid state transformations in wet granules. Even small amounts of a non-expected
polymorph can be detected.
NIR spectroscopy
To investigate if the same conclusions can be made when analyzing the granules with NIR
spectroscopy, runs 6 till 19 were taken into account (runs 1 till 5 were not taken into account, since
inadequate NIR spectra were obtained due to sampling problems). PCA was performed on the
complete spectral range (10000 to 4500 cm-1) resulting in three principal components. These
Gewijzigde veldcode
explained 97.1 % of the variation. PC 1 explained 69.11 % of the spectral variation, PC 2 18.82% and
PC 3 9.13 %. In order to make a distinct classification possible, secondary observation identifiers (i.e.,
assignment of colors, which are corresponding to certain process settings, to the scores) were added
to all spectra.
The first principal component differentiates the granulation experiments after water content (Figure
8a). Granules produced with a high liquid amount are spread along the positive part of the PC 1-axis,
while those produced with a low liquid amount can be found in the negative part of PC 1. This is
confirmed by the loadings of this first principal component, which clearly shows two water bands
(Figure 8b). The spectral band at 5222 cm-1 corresponds to the OH-stretching and bending vibrations
of water molecules, while the 7035 cm-1 band can be attributed to first overtone OH-stretching
vibrations. Two more vibration are contributing to these peaks, resulting in two shoulders at 6880
cm-1 and 5168 cm-1. These two bands are caused by lactose monohydrate [33].
In Figure 9a, the PC 2 versus PC 3 scores plot was plotted and data were labeled and examined after
secondary observation identifiers. Most spectra of the granules produced with a low concentration
of granulation liquid can be found in the positive part of the PC 3-axis. The same trend can be seen
for granules produced in a warm granulation barrel (Figure 9b). Classifying data after powder feed
rate or screw speed did not result in any clear clustering.
The loadings plot of PC 3 (Figure 9c) shows an absorption band of OH vibrations at 7035 cm-1. This
water-bond is oppositely directed, compared to the bands at 5947 cm-1 and 6009 cm-1, which are
specific peaks for anhydrous theophylline. Furthermore, a remarkable peak at 5168 cm-1 can be
attributed to lactose monohydrate, which has the same concentration in all granules [33]. NIR
confirms hence the Raman analysis findings. The granulation experiments corresponding to the
scores in the positive part of the PC3-axis most likely contain remaining anhydrous theophylline. NIR
only shows the variation in solid state, when different barrel temperatures or liquid concentrations
are applied.
An overview of the runs containing an amount of anhydrous theophylline is given in Table 3. Runs (6-
19) which were thought to contain anhydrous theophylline when investigated with Raman are all
confirmed by NIR.
Conclusions
This study gave an insight in the possibilities of implementing PAT-tools in a continuous twin-
screw granulation process. Both Raman and NIR showed to be appropriate tools for understanding
the solid state behavior of theophylline during wet granulation. Peak differentiation and polymorph
differentiation was more definite in Raman spectra than in NIR spectra.
The In-Line Particle Probe showed to be a promising tool for the in-line measurement of particle size
distribution. More investigation on the measurement technique will be necessary to obtain
adequate measurements of wet granules. The main challenge is to avoid fouling of the optical
surfaces.
Performing a DOE, allowed to reveal the most influencing granulation parameters upon granule size
and the solid state of the API. The temperature of the barrel and the added amount of granulation
liquid, and to a lesser extent the powder feed rate showed to have an effect upon the solid state of
the wet granules. The powder feed rate and the barrel temperature had significant effects upon the
granule size.
This paper proves that at-line measurements of both solid state and particle size during a continuous
twin-screw granulation process are feasible. Given these promising results of the presented tools,
the next step will be the design of a new process analyzer interfacing device. This device should
permit integration of PAT-tools in the ConsiGma™-25, allowing them to provide real-time
information. Future studies will focus on the use of the in-line measured critical process and product
information to steer the next process steps (e.g., feed forward adjustment of process settings of
further process steps).
Acknowledgments
Mr. Gerrit Fonteyne Jr. is gratefully acknowledged for the design and production of the Megacuvet.
The authors want to thank Pharm. Andre da Fonseca Antunes for technical assistance and are
grateful to Ms. Anneleen Burggraeve for experimental support.
Declaration of interest
The authors report no declarations of interest.
References
1. Leuenberger, H., New trends in the production of pharmaceutical granules: batch versus continuous processing. European Journal of Pharmaceutics and Biopharmaceutics, 2001. 52(3): p. 289-296.
2. Levin, M., Introduction, in Pharmaceutical Process Scale-up2001, Marcel Dekker: New York. p. vii-x.
3. Leuenberger, H., Scale-up in the 4th dimension in the field of granulation and drying or how to avoid classical scale-up. Powder Technology, 2003. 130(1-3): p. 225-230.
4. FDA-Adminstration. Guidance for Industry - PAT - A Framework for Innovative Pharmaceutical Development, Manufacturing, and Quality Assurance. 2004; Available from: http://www.fda.gov/downloads/Drugs/GuidanceComplianceRegulatoryInformation/Guidances/ucm070305.pdf.
5. Guidance for Industry - Process Validation: General Principles and Practices - January 2011. 2011; Available from: http://www.fda.gov/downloads/Drugs/GuidanceComplianceRegulatoryInformation/Guidances/UCM070336.pdf.
6. International Conference on Harmonisation of Technicl Requirements for Registration of Pharmaceuticals for Human Use. Pharmaceutical Development Q8. 2009 [cited 2011; Available from: http://www.ich.org/fileadmin/Public_Web_Site/ICH_Products/Guidelines/Quality/Q8_R1/Step4/Q8_R2_Guideline.pdf.
7. International Conference on Harmonisation of Technical Requirements of Regeistration of Pharmaceuticals for Human Use. Quality Risk Management Q9. 2005 [cited 2011; Available from: http://www.ich.org/fileadmin/Public_Web_Site/ICH_Products/Guidelines/Quality/Q9/Step4/Q9_Guideline.pdf.
8. International Conference on Harmonisation of Technical Requirements for Registration of Pharmaceuticals for Human Use. Pharmaceutical Quality System Q10. 2008 [cited 2011; Available from: http://www.ich.org/fileadmin/Public_Web_Site/ICH_Products/Guidelines/Quality/Q10/Step4/Q10_Guideline.pdf.
9. Naelapaa, K., et al., Building quality into a coating process. Pharmaceutical Development and Technology, 2010. 15(1): p. 35-45.
10. Gamlen, M.J. and C. Eardley, CONTINUOUS EXTRUSION USING A BAKER PERKINS MP50 (MULTIPURPOSE) EXTRUDER. Drug Development and Industrial Pharmacy, 1986. 12(11-13): p. 1701-1713.
11. Lindberg, N.O., C. Tufvesson, and L. Olbjer, EXTRUSION OF AN EFFERVESCENT GRANULATION WITH A TWIN SCREW EXTRUDER, BAKER PERKINS MPF 50-D. Drug Development and Industrial Pharmacy, 1987. 13(9-11): p. 1891-1913.
12. Werani, J., et al., Semicontinuous granulation - the process of choice for the production of pharmaceutical granules? Powder Technology, 2004. 140(3): p. 163-168.
13. Betz, G., P. Junker-Burgin, and H. Leuenberger, Batch and continuous processing in the production of pharmaceutical granules. Pharmaceutical Development and Technology, 2003. 8(3): p. 289-297.
14. Vervaet, C. and J.P. Remon, Continuous Granulation, in Handbook of Pharmaceutical Granulation Technology, D.M. Parikh, Editor 2009, Informa Healthcare: New York. p. p. 308-322.
15. Vervaet, C. and J.P. Remon, Continuous granulation in the pharmaceutical industry. Chemical Engineering Science, 2005. 60(14): p. 3949-3957.
16. Keleb, E.I., et al., Continuous twin screw extrusion for the wet granulation of lactose. International Journal of Pharmaceutics, 2002. 239(1-2): p. 69-80.
17. Keleb, E.I., et al., Twin screw granulation as a simple and efficient tool for continuous wet granulation. International Journal of Pharmaceutics, 2004. 273(1-2): p. 183-194.
18. Van Melkebeke, B., C. Vervaet, and J.P. Remon, Validation of a continuous granulation process using a twin-screw extruder. International Journal of Pharmaceutics, 2008. 356(1-2): p. 224-230.
19. Djuric, D., et al., Comparison of two twin-screw extruders for continuous granulation. European Journal of Pharmaceutics and Biopharmaceutics, 2009. 71(1): p. 155-160.
20. Aulton, M.E. and B. M., Fluidised bed granulation—factors influencing the quality of the product. Int. J. Pharm. Technol. Prod. Manuf., 1981. 2(4): p. 24-29.
21. Kristensen, H.G. and T. Schaefer, GRANULATION - A REVIEW OF PHARMACEUTICAL WET-GRANULATION. Drug Development and Industrial Pharmacy, 1987. 13(4-5): p. 803-872.
22. Eriksson, L., et al., Design of Experiments - Principles and Applications2008, Umea: Umetrics. 23. Suzuki, E., K. Shimomura, and K. Sekiguchi, THERMOCHEMICAL STUDY OF THEOPHYLLINE
AND ITS HYDRATE. Chemical & Pharmaceutical Bulletin, 1989. 37(2): p. 493-497. 24. Heinz, A., et al., Analysis of solid-state transformations of pharmaceutical compounds using
vibrational spectroscopy. Journal of Pharmacy and Pharmacology, 2009. 61(8): p. 971-988. 25. Bauer, J., et al., Ritonavir: An extraordinary example of conformational polymorphism.
Pharmaceutical Research, 2001. 18(6): p. 859-866. 26. Airaksinen, S., et al., Effects of excipients on hydrate formation in wet masses containing
theophylline. Journal of Pharmaceutical Sciences, 2003. 92(3): p. 516-528. 27. Wikstrom, H., P.J. Marsac, and L.S. Taylor, In-line monitoring of hydrate formation during wet
granulation using Raman spectroscopy. Journal of Pharmaceutical Sciences, 2005. 94(1): p. 209-219.
28. Burggraeve, A., et al., Evaluation of in-line spatial filter velocimetry as PAT monitoring tool for particle growth during fluid bed granulation. European Journal of Pharmaceutics and Biopharmaceutics, 2010. 76(1): p. 138-146.
29. Jorgensen, A., et al., Hydrate formation during wet granulation studied by spectroscopic methods and multivariate analysis. Pharmaceutical Research, 2002. 19(9): p. 1285-1291.
30. Ticehurst, M.D., R.A. Storey, and C. Watt, Application of slurry bridging experiments at controlled water activities to predict the solid-state conversion between anhydrous and hydrated forms using theophylline as a model drug. International Journal of Pharmaceutics, 2002. 247(1-2): p. 1-10.
31. Yoshihashi, Y., et al., Determination of heat of hydration and hydration kinetics of theophylline by thermal analysis. Chemical & Pharmaceutical Bulletin, 1998. 46(7): p. 1148-1152.
32. Zhang, G.G.Z., et al., Phase transformation considerations during process development and manufacture of solid oral dosage forms. Advanced Drug Delivery Reviews, 2004. 56(3): p. 371-390.
33. Ali, H.R.H., H.G.M. Edwards, and I.J. Scowen, Noninvasive in situ identification and band assignments of some pharmaceutical excipients inside USP vials with FT-near-infrared spectroscopy. Spectrochimica Acta Part a-Molecular and Biomolecular Spectroscopy, 2009. 72(4): p. 890-896.
Tables
Equipment Variables Product Variables Process Variables Starting Powder Properties Liquid Characteristics Screw configuration Particle size distribution Viscosity Powder feed rate Screw length Homogeneity of raw material powders Surface tension Bridge breaker speed Length- to-diameter ratio of the screws Presence of agglomerates in premix Solid-liquid tension Filling degree of barrel Diameter of the granulation liquid inlet nozzels Segregation during feeding
Screw speed
Powder density
Barrel temperature
Powder feeding method Cohesion Binder type Liquid feed rate Place of liquid addition into barrel Particle shape Binder concentration
Table 2. Overview of the design experiments – Calculated D50 values obtained after two process minutes.
Run Water Content Barrel T Powder Feed Rate Raman NIR
Dry - (8,38%) Hot - (40 °C) Full - (25kg/h)
1 X 2 X X +++ 3 X +++ 4 5 X X +++ 6 7
8 X X X +++ +++ 9 X X +++ +++
10 X X 11 X 12 X 13
X
14 X 15
16 X
X +++ +++ 17
18 X X +++ +++ 19 X X X +++ +++
Table 3. Overview of all DOE granulation experiments. Both significant parameters and spectroscopic responses are listed. Runs in which Raman and/or NIR-spectroscopy show the presence of anhydrous theophylline are indicated with +++.